RE: 5.2

john_slatin writes:
 > 
 > I see your point, Lee, and it makes sense.  I think we would be wise to
 > avoid phrases like "user agent engines" because the concept of an "engine"
 > is foreign to most people.  The ones who recognize the term might associate
 > it with "search engines," but they still won't know what the word "engine"
 > means in that context, either.

Ultimately, I don't think it is a communication issue. Rather it is an
attempt to tie the checkpoint too much to the structure of those
implementations in which the concept of an "engine" makes sense. I
suggest we return to the underlying issue: not only is there a variety
of user agents, but people with particular needs addressed by the
guidelines have reason to select implementations that best meet their
needs, including custom-developed user agents. Content that relies on
a single, specific implementation of the formats, API's and protocols
it uses, will be inaccessible to users who, for whatever reasons, find
that their needs are best met, or can only be met, by a different
implementation from that on which the author has chosen to depend.

The current discussion relates only to how the requirement should be
expressed; there is at present no dispute as to its importance or
place in the guidelines, issues which I therefore leave aside in what
follows. I think the simplest and most accurate mode of expression is
to say that there exist multiple, independent and interoperable implementations of the
technologies (i.e., formats, protocols, API's etc.) used by the
content.

This is different from saying that the content is implemented in
multiple user agents, in as much as it is here explicitly prescribed
that the implementations must be independent, provided that a suitable
definition of independence is given, e.g., that they have not been
developed by the same entity, or that they don't have substantial code in
common, or whatever seems most reasonable as a requirement.

Received on Thursday, 26 December 2002 01:18:15 UTC